The Money Leaving the Building

A guy I have known for years spent most of last spring fighting for a price increase.

Eleven percent. He agonized over it for six weeks. Rewrote the email nine times. Lost sleep. Ran the numbers on how many clients he could afford to lose. Finally sent it, held his breath, and got one cancellation out of forty two.

He was thrilled. It was a real win.

Then in July he exported twelve months of card statements for his accountant and found 31 active software subscriptions. He could confidently explain what nine of them did. Four of them were tools he had personally stopped using more than a year ago. Two were duplicates of each other bought by different people six weeks apart. One was a $340 a month analytics platform that nobody in the company had logged into since the previous September.

The annual total on the stuff he couldn’t justify was just under 26,000 dollars.

He had spent six weeks of emotional energy fighting for a price increase worth less than the money quietly walking out his own back door every year.

I’m not telling that story to make him look bad. I’m telling it because it’s the most common blind spot I see in businesses that are otherwise run tight. We audit revenue obsessively. We look at close rates, pipeline velocity, average deal size, churn. And almost none of us ever runs the same discipline on the other side of the ledger.

Today we fix that. Two hours, one sitting.

It isn’t carelessness. It’s structural.

Adding a subscription has a trigger, an owner, and a moment. Somebody needs something, somebody buys it, there’s a decision and a date.

Removing one has none of that. There’s no moment when a tool announces it has become useless. It just quietly stops being opened while the card keeps getting charged. Nobody’s job description includes noticing.

The numbers on this got genuinely absurd over the last three years. The average ten person team now runs 28 active software subscriptions. Three years ago that number was 19. Small and mid sized businesses waste roughly 27 percent of their software budget on tools that are unused, underused, or duplicated. Two thirds of businesses auto renewed at least one tool in the last twelve months that they no longer use.

And the fastest growing category of waste is exactly the one you’ve been adding to hardest. AI tools carry the highest waste rate of any software category, somewhere north of 40 percent, partly because everybody signed up for six of them in a panic during the last two years and partly because consumption based pricing means the bill is a surprise instead of a decision.

The good news is the fix isn’t complicated. Businesses that run a real audit for the first time recover an average of around 6,800 dollars a year. For most of the men reading this it’ll be considerably more, because you’re the one with the company card and you’re the one who signed up for things at 11pm.

Block the time. Actually block it. This isn’t a task you fit between calls.

Don’t do this from memory. Your memory is the reason the problem exists.

Three sources, all of them:

Export twelve months of transactions from every business card and bank account. Sort by merchant. Anything that appears more than twice is a subscription whether you thought of it that way or not.

Search your email inbox for “receipt,” “your invoice,” “subscription renewal,” “your plan,” and “payment confirmation.” Set the date range to the last twelve months. You’ll find things that never hit the card you were looking at.

Check your Apple and Google account subscription pages directly. There’s almost always something living there that never shows up in a normal accounting review because it bills as a single anonymous line item.

Now you have a list. It will be longer than you expected. That’s fine and it’s the point.

One row per tool. Four columns.

Tool name. Annual cost, which means monthly times twelve, because monthly pricing hides the real number and that’s exactly why it exists. Last time it was genuinely used, and if you aren’t sure, that’s your answer. Owner, meaning the specific human being who would notice tomorrow if it vanished.

That fourth column does most of the work. Any tool where the owner is “the company” or “us” is a tool with no owner, and a tool with no owner is a tool nobody will defend, which usually means nobody needs it.

Load bearing. Remove it and something breaks today. Your CRM, your accounting, your payment processing, your email platform. Usually between four and eight tools in a small business. Fewer than you assume.

Redundant. You’re paying two or three companies to do the same job. Two project trackers because a team migrated halfway. Three places files live. A scheduler that came with your CRM plus a standalone scheduler nobody turned off. This bucket is almost always bigger than people expect.

Nostalgic. You’re paying for a version of your business that no longer exists. The tool you bought when you were going to launch that thing. The platform for the service line you sunset in 2024. The seats for the team you had before you restructured. These are the hardest to cut because canceling feels like admitting a plan died, and it did, and the subscription isn’t going to bring it back.

Zombie. Nobody has opened it in 90 days and nobody can explain why you have it. Frequently a trial that converted, a tool a departed employee bought, or something you signed up for after a conference.

Zombies die today. Not next week. Today, while you have the list open, because a decision deferred is a decision to keep paying.

Nostalgic tools get a 30 day proof window. Put it in writing: if this isn’t used meaningfully by October 16, it goes. Then actually cancel on October 16. The window isn’t a stay of execution, it’s a test, and most of them fail it.

Redundant tools require a choice, and the choice needs a name attached and a date. Pick the winner, set a migration deadline inside 14 days, cancel the loser on that date whether the migration finished or not. If you don’t set the cancellation date first, the migration will take four months and you’ll pay for both the entire time. I have watched this happen more times than I can count.

Load bearing tools don’t get canceled. They get renegotiated, which almost nobody does.

This is the step people skip, and it’s free money.

Software companies price for churn. They have a retention team with real authority to discount, and that authority is almost never exercised because customers don’t ask. Here’s the email. Send it 45 days before renewal, not on renewal day, because on renewal day you have no leverage.

“Hi [name], our renewal is coming up on [date] and I’m doing a full review of our stack this quarter. We’re currently on [plan] at [price] for [number] seats. Three questions before I make a recommendation internally. What’s the discount for paying annually up front? What does the pricing look like at [lower number] seats, since our actual active usage is [real number]? And is there a tier below ours that still includes [the two features we actually use]? I’d like to keep working with you. I just need the number to make sense against what we’re actually using.”

That’s it. No threats, no drama. Three specific asks, one honest reason.

In my experience about a third of vendors come back with something meaningful, usually 15 to 25 percent, sometimes a full tier downgrade you didn’t know existed. On a stack costing 4,000 a month that’s worth having a boring afternoon over.

The line about actual active usage is the one that moves them, so know the number before you send it. Almost every platform has an admin page showing last login by user. Pull it first.

Separate exercise, five minutes, frequently the biggest single line.

Pull the user list on every multi seat tool you own. Compare it against your actual current team.

You’ll find seats for people who left. You’ll find seats for contractors who finished a project in February. You’ll find a tool sized for the ten person team you had when you bought it and you now have six.

Here’s the thing that makes this invisible: the invoice doesn’t change when someone stops using a seat. It looks exactly the same as it did when everyone was active. Nothing alerts you. A ten seat plan serving six people is wasting 40 percent every single month and the bill gives you no signal at all.

Do this quarterly and put it on the calendar as a recurring event with your name on it.

Give this its own section on your ledger, because it doesn’t behave like the rest of the stack.

Most of us signed up for a pile of AI tools over the last twenty four months, several of them at twenty bucks a pop, some of them at two hundred. They individually feel too small to bother reviewing, which is exactly how they add up to a serious number without ever being a decision.

Three moves.

Consolidate. Most men are paying four separate subscriptions for capabilities that substantially overlap. Pick the one you actually reach for and cut the rest. You won’t miss them and you know it. And if you genuinely use several models for different jobs, which some of us do, the cheaper structure is one account that gets you all of them rather than four separate twenty dollar charges. I run Galaxy for that reason alone. One line on the statement instead of five, and I stopped relitigating which subscription to cancel every time a new model shipped.

Cap the consumption ones. Anything billed on usage instead of a flat seat needs a hard budget limit set inside the platform, not a good intention. Surprise charges on consumption pricing are now one of the most common line item shocks in a monthly close.

Convert the good ones into systems. This is the part almost nobody does. A tool you open manually saves you minutes. A tool wired into a workflow that runs whether or not you remember to open it saves you hours.

That last one is where consolidation actually pays. I have replaced three separate subscriptions in my own stack with a single Make scenario more than once. Form comes in, record gets created, notification fires, follow up sequence starts, all of it running without a human. What used to be three tools and a person remembering to check something is now one line on a bill. The point isn’t the automation for its own sake. The point is that every workflow you wire together is a subscription you get to stop paying for and a task you get to stop remembering.

Everything you decided to keep gets a calendar entry 45 days before its renewal date. In the entry, put the owner’s name, the current annual cost, and one question:

“Would we buy this again today, at this price, knowing what we know?”

That’s the entire system. It takes an hour to set up once and it permanently ends the problem of being surprised by a charge for something you meant to cancel in April.

Same audit, four more categories, and honestly some of these are bigger.

Contractors and retainers. The monthly retainer that made sense when the scope was three times larger. The agency you kept because switching sounded exhausting. Look at deliverables received in the last 90 days against the invoice for those 90 days.

Memberships and sponsorships. The mastermind you haven’t attended since March. The trade association nobody in the company has used. The conference sponsorship you renewed because a rep called you in a good mood.

Merchant processing. This is the sleeper. If you’ve been on the same processing agreement for more than two years, you’re almost certainly overpaying by somewhere between 20 and 50 basis points, and on real volume that’s a serious annual number. Get two competing quotes and take them to your current processor. This one call is frequently worth more than the entire software audit.

Insurance and banking. Same principle. Bundled policies drift. Bank fees drift. Nobody rebids them because nothing forces it.

Here’s the reframe, and I want you to actually run this number on your own business.

Say you find 1,200 a month. Twelve of those are worth 14,400 a year, which sounds fine, sounds like a nice little win, sounds like something you’d mention in passing.

Now check your net margin. If you run at 10 percent, that 14,400 in saved cost is the exact equivalent of generating 144,000 dollars in new revenue.

A hundred and forty four thousand dollars.

Ask yourself how many hours you’d put into a plan to add 144,000 in new revenue this quarter. How many calls. How much ad spend. How much of your own attention.

Now ask how many hours you were about to put into two hours with a bank statement and a spreadsheet.

That’s the trade. It’s the least glamorous work in the business and it’s very close to the highest hourly rate available to you between now and December 31.

Nobody’s going to congratulate you for it. There’s no story in it, nothing to post about, no moment where somebody shakes your hand. It’s just two boring hours that quietly make the rest of the year worth more.

Do it this week, before the Q1 renewals stack up in January and you’re paying for another twelve months of something you meant to kill in September.

Refined. Relentless. Unapologetic.

Marcus